While it's not known exactly how this is controlled, the idea circulating has always been the levels of tissue M-CSF, GM-CSF and IL-34. Macrophages need growth factors to survive, and consume these at a constant rate. The levels of these factors is tightly regulated in homeostasis so it sort of works to equilibrium. I should also point out that many tissues do not have continual monocyte recruitment and the macrophages in these homeostatic tissues are self-replenishing (or try to be at least). Most of these studies are murine, so we have yet to conclude on human tissue macrophages.
However in inflammation when macrophage numbers drop, the tissue M-CSF will theoretically increase during the resolution, this may drive macrophage proliferation to repopulate the niche, or allow monocytes to fill it. Additionally, IL-4 can increase these numbers during parasite infections/ lung allergy, and monocytes are recruited via CCR2, so it is not just limited to growth factors in variable organism settings.
This subject has been recently reviewed by Martin Guilliams and Charlotte Scott.
https://www.nature.com/articles/nri.2017.42
What is important here, is that the immune system doesn't control this, the tissue does. The term immune system is pretty broad though so its a little subjective. I don't really consider tissue-resident macrophages to be part of the immune system any more, (well, as much as skin epithelia aren't a part of it). These cells contribute to immune protection and surveillance, but aren't part of a systemic system. So either you say that pretty much the whole organism is part of the immune system, or seal it off to a systemic system of monocytes/inflammation associated macrophages, neutrophils, t-cells etc. It's entirely subjective and I have no problem with any interpretation.
This is a very cool question. I spent about ten years nearly forty years ago trying to address it in synovial tissue and my conclusion was that it is really very complicated.
The soluble factor effects Luke mentions are clearly important and I think we have made significant progress since the days of van Furth and Spector on bone marrow derived and locally derived populations.
However, a lot of tissue actually behave as two dimensional rather than three dimensional functional units - synovial intima being a good example. So volume is not relevant. Macrophages line up on the surface of synovium like fried eggs on a pizza fiorentina. This is probably in part due to the specialised expression of very high level VCAM-1 by their neighbour synovial fibroblasts. But we also had reason to think that both cells 'know where they are' because of a mesh of fibrillin-1 laid down on the tissue surface. The fibrin-1 fibres are painted with various molecules including cDAF (CD55). Similar but subtly different situations occur around muscle cells within muscle, in meninges and in pericardium. It is almost as if at some time in the past some traffic warden cells have painted the matrix with white and yellow lines to tell the macrophages not only where to park but also what to express (e.g. CD16, CR3).
In inflammation some of the rules seem to change and others stay the same. New macrophage parking areas are set up in association with lymphocytes and also dense collagen zones (where the haemosiderin-laden ones build up).
I suspect there is still room for several lifetimes work on this. The main lesson I learnt is that cells are very intelligent things. They make very complex decisions about where to be and whether or not to divide.